DNA Based Computingによる点群データの外形推定

抄録

This study deals with a computational method called DNA Based Computing (DBC) that underlies the biological phenomenon called “central dogma.” It (central dogma) means that once the information of DNA/RNA has passed into proteins (sequences of Amino Acids) it cannot get out again. Proposed DBC adopts a set of user-defined rules to generate a DNA-array from given problem-relevant information (e.g., image, sensory input). Next, mRNA-arrays are transcripted from DNA-array with user-defined conversion rules. Afterward, the Amino Acids (AAs)-arrays (or proteins) are synthesized from the mRNA-array according to genetic codes known as Codons. Finally, a set of rules to solve a problem is extracted from the AAs-array or protein. These rules can be based on the informational characteristics of the AAs (entropy, absence, presence, abundance, and frequency of some selected amino acids, and relationships among their likelihoods). In this article, we apply the DBC to determine the outer boundary (concave-hull) of a complex point-cloud generated by the IFS fractals. We show the computational performance (e.g., iteration count, the accuracy of the outline, and the amount of information) by numerical experiments. Two types of rules are considered for DNA generation (ACGT rule, ACTG rule) and mRNA transcription (asymmetric and symmetric rule), respectively. We find that the proposed method can successfully determine the desired outer boundary when a critical number of points populates given point-cloud.